{"id":"W3042470307","doi":"10.48550/arxiv.2007.06731","title":"Regularized linear autoencoders recover the principal components, eventually","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Model Reduction and Neural Networks","field":"Physics and Astronomy","cited_by":11,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Regularization (linguistics); Simple (philosophy); Feature learning; Representation (politics); Principal component analysis; Computer science; Gradient descent; Mathematics; Dimension (graph theory); Artificial neural network; Convergence (economics); Mathematical optimization; Applied mathematics; Artificial intelligence; Algorithm; Combinatorics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0001243255,0.0003315009,0.0003384265,0.00005468755,0.0002326075,0.00005750084,0.0007272292,0.0001571818,0.00068467],"category_scores_gemma":[0.000005467533,0.0002955144,0.0004137733,0.0002301593,0.0001313523,0.0001009262,0.0008098877,0.001048782,0.0001482712],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007386274,"about_ca_system_score_gemma":0.0001411495,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002085973,"about_ca_topic_score_gemma":0.000005522906,"domain_scores_codex":[0.9983923,0.0001928748,0.0002283941,0.0007693383,0.0001047158,0.0003124017],"domain_scores_gemma":[0.9987327,0.00006263799,0.0002704731,0.000671634,0.00007525994,0.0001872319],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002571113,0.0001571393,0.0020268,0.00003462849,0.0005059405,0.00003896388,0.0002695803,0.8853534,0.0001324199,0.1056791,0.004906116,0.0006387615],"study_design_scores_gemma":[0.0007797133,0.00002874403,0.0005288887,0.0000587556,0.0001851693,9.16577e-7,0.0001758188,0.9516307,0.00006542779,0.03091273,0.01519566,0.0004374738],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6094493,0.00002956537,0.3726154,0.001646654,0.001829294,0.000914658,0.0001097528,0.000238319,0.0131671],"genre_scores_gemma":[0.9920166,0.00003818679,0.0003173818,0.0001841809,0.0006554255,0.000002227615,0.000173664,0.00003290153,0.006579482],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3825673,"threshold_uncertainty_score":0.9999497,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1089098758402249,"score_gpt":0.2025313578770978,"score_spread":0.09362148203687297,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}